Universidade de Aveiro Ano 2018
Departamento de Eletrónica Telecomunicações e Informática
Miguel Pais Machado
Visible Light Communications for Drivers Assistance
Systems
Sistemas de Comunicação por Luz Visível Aplicados
para Assistência ao Tráfego Automóvel
Universidade de Aveiro Ano 2018
Departamento de Eletrónica Telecomunicações e Informática
Miguel Pais Machado
Visible Light Communications for Drivers Assistance
Systems
Sistemas de Comunicação por Luz Visível Aplicados
para Assistência ao Tráfego Automóvel
Dissertação apresentada à Universidade de Aveiro para cumprimento dos requisitos necessários à obtenção do grau de Mestre em
Engenharia Eletrónica e Telecomunicações, realizada sob a orientação científica do Doutor Luis Nero Alves, Professor Auxiliar do
Departamento de Eletrónica, Telecomunicações e Informática da Universidade de Aveiro
o júri
Presidente Prof. Doutora Susana Isabel Barreto de Miranda Sargento Professora Associada C/ Agregação do Departamento de Engenharia Eletrónica, Telecomunicações e Informática da Universidade de Aveiro
Orientador Prof. Doutor Luis Filipe Mesquita Nero Moreira Alves
Professor Auxiliar do Departamento de Engenharia Eletrónica, Telecomunicações e Informática da Universidade de Aveiro
Arguente Prof. Doutora Mónica Jorge Carvalho Figueiredo
Professora Adjunta do Departamento de Engenharia Eletrotécnica da Escola Superior de Tecnologia e Gestão do Instituto Politécnico de Leiria
Acknowledgements The proposed topic was a chance to develop new knowledge about optical wireless communications, contribute to an ongoing project as well as to synthetize and review innumerable aspects and subjects lectured along the master’s course. This dissertation covers my final work as a student of University of Aveiro, and more singularly marks the end to an era of studies on the master’s course on Electronical and Telecommunications Engineering.
It is now time for a new adventure outside the academic world, but I shall cherish everyone that have helped me along this path. Without disregard to anyone that I may forget to mention, here goes the owed thanks in the development of this work:
To my supervisor Professor Luis Filipe Mesquita Nero Moreira Alves, for the continuous support and belief.
To all my colleagues in the laboratory for integrated circuits, Luis Abade, Luis Rodrigues, Debarati, Shusmitha, João, André, Patricia, Miguel and other’s that contribute to the group, for a great work environment, for being a good source of information and discussion.
To the Telecommunications institute of Aveiro (IT), for providing the facilities and resources.
To all the teachers along the way, old and new.
To my parents, Maria Isabel da Costa Pais and José Carlos Soares Machado e Silva, for all the love and patience they had with me and to all friends who had the ability of understanding the difficult plight the last year has been.
palavras-chave I2C, C2C, linha de visão, radiação difusa, relação sinal-ruido, taxa de erro, fontes de ruido
resumo Motivado em promover o tópico de segurança rodoviária e sistemas de informação, este trabalho providência um estudo dedicado a sistemas de comunicação por luz visível (VLC) para aplicação em cenários de exterior. O tópico desenvolvido faz parte de sistemas de transporte inteligentes (ITS) cujo propósito é a disseminação de sistemas de segurança no tráfego e transferência de informação, para aplicações de segurança. A tecnologia VLC aplicada a sistemas de comunicação de tráfego rodoviário suscita elevado interesse devido a vantagens que esta apresenta. O uso de LED’s em semáforos e faróis de carros começa a ser bastante comum. Com a combinação de diferentes valências, como iluminação e transferência de dados no mesmo dispositivo, a tecnologia VLC torna-se muito atrativa para a implementação em sistema de comunicação exterior dedicados a sistemas de informação e controlo de tráfego.
O canal de comunicação VLC exterior apresenta condições variáveis, devido ao fato de existirem condições ambientais diferentes. Um grave problema neste tipo de canal de comunicação é a presença de ruido Shot, que é normalmente gerado devido á radiância causada por diferentes fontes de luz de fundo.
Nesta dissertação estão presentes dois tipos de cenários para sistemas de informação de tráfego, em que o primeiro dedica-se á comunicação semáforo-carro (I2C) e o segundo cenário para a comunicação entre carros (C2C). Para simular o desempenho do canal de comunicação com diferentes condições ambientais, foram implementados em MATLAB modelos para a propagação ótica, descrição do emissor, recetor e fontes de ruido. Também foram incluídos modelos para diferentes fontes óticas de radiação, com medições de campo da iluminância incidente num foto recetor e modulado o impacto na geração de ruido.
Nas simulações de desempenho da comunicação por luz visível, foram considerados diferentes esquemas de modulação da informação com o
keywords I2C, C2C, line-of-sight, diffuse radiation, signal-to-noise, bit-error-rate, optical noise
abstract Motivated by the topic of promoting traffic safety applications and information systems, this work aims to bring a study on VLC outdoor application scenarios. The developed topic is part of intelligent transportation systems (ITS) that aim at the delivery of traffic safety and information amongst other safety functions. VLC technology in traffic communication applications gains interest due to some advantages it presents. The use of LEDs in traffic signaling infrastructures and vehicle headlights started to be a growing standard. With the combination of illuminating proprieties and communication in the same device, VLC becomes a very attractive technology for the implementation of outdoor communication systems for traffic information and control.
Outdoor VLC channels present variable ambient conditions, with the presence of different optical sources. One major problem in this communication channel is the presence of shot-noise, generated by optical background radiance from different light sources.
This dissertation presents two different communication scenarios for traffic information systems, the first being directed at the infrastructure to car (I2C) link and the second one for car to car (C2C) communication. In order to simulate the communication link performance with variable ambient channel conditions, several models for optical propagation, emitter, receiver and noise sources were implemented in MATLAB. Models for different optical sources were also implemented, with field measurements on the illuminance incident on a photo detector and their impact on the noise generated.
In the simulation’s performance of the VLC link, several baseband modulation schemes were considered, aiming at the assessment of link performance, based on the traditional digital modulation performance metrics.
Contents
List of Figures ... v List of Tables ... ix Acronyms ... xi 1. Chapter 1 ... 1 1.1 Introduction ... 1 1.2 Motivation ... 3 1.2.1 VLC Applications ... 4 1.3 Objectives ... 5 1.4 Dissertation structure ... 5 2. Chapter 2 ... 72.1 System Model Description ... 7
2.2 Channel Modelling ... 8
2.2.1 Emitter technical characteristics ... 8
2.2.2 Emitter Model ... 12
2.2.3 Traffic Light Emitter Matrix ... 14
2.2.4 Receiver Model ... 18
2.2.5 I2C LOS Model Description ... 21
2.2.6 Optical Power Path- I2C ... 23
2.3 Channel Noise Sources ... 24
2.3.1 Electronics Noise ... 25
2.3.2 Signal and Noise ... 27
2.3.3 Black Body Radiation ... 28
3.1.1 Channel Bandwidth Limitations ... 32
3.2 Digital Baseband Modulations ... 36
3.2.1 IIM – Impulse Intensity Modulation – DD – Direct Detection... 38
3.2.2 PPM Modulation ... 41
3.2.3 DPIM Modulation ... 43
3.2.4 DH-PIM Modulation ... 44
3.3 Modulations Comparison ... 47
4. Chapter 4 ... 49
4.1 Variable Ambient Optical Conditions on System Performance ... 49
4.2 Ambient Noise Characterization ... 50
4.2.1 Low optical natural light ... 50
4.2.2 Artificial light ... 54
4.2.3 High intensity natural light ... 58
4.3 Noise Reduction ... 62
5. Chapter 5 ... 65
5.1 C2C Communication link ... 65
5.2 Emitter model ... 66
5.2.1 Radiation pattern ... 68
5.3 LOS and NLOS propagation ... 70
5.3.1 Platform illustration ... 70
5.4 Multipath Propagation ... 73
5.4.1 Channel impulse response ... 73
5.4.2 Delay Spread ... 74
5.5 C2C Modulations Performance... 76
6. Chapter 6 ... 81
6.1 Conclusion ... 81
Bibliography ... 83 Appendix A ... 87
List of Figures
Figure 1.1 Electromagnetic Spectrum ... 1
Figure 1.2 Claudde's Chappe optical telegraph ... 2
Figure 2.1 I2C scenario ... 7
Figure 2.2 Electro optical LED efficiency ... 10
Figure 2.3 Normalized human eye spectral sensitivity curve and red LED spectral power ... 11
Figure 2.4 LED emission perpendicular to the junction plane ... 12
Figure 2.5 Polar plots of Lambertian radiation patterns ... 13
Figure 2.6 LED matrix (200mm) ... 14
Figure 2.7 Emitting plane of discrete LEDs with circular ring arrangement ... 15
Figure 2.8 Matrix luminous intensity vs LED driving current ... 17
Figure 2.9 FDS100 photodiode spectral responsivity ... 18
Figure 2.10 Optical band-pass filter transmittance response ... 19
Figure 2.11 Receiver model ... 20
Figure 2.12 Average optical power received (dBm) ... 24
Figure 2.13 Power spectrum of a constant photocurrent on a photodetector and the noise spectral distribution ... 26
Figure 2.14 Black body emission spectrum varying with temperature ... 28
Figure 2.15 Normalized PSD from natural and artificial optical sources ... 30
Figure 3.1 Voltage mode emitter circuit ... 33
Figure 3.2 Receiver circuit ... 34
Figure 3.3 Measured optical bandwidth of OVLFR3C7 ... 35
Figure 3.4 Typical VLC block diagram ... 36
Figure 3.5 Rectangular pulse shape and PSD of the Fourier transform ... 37
Figure 3.6 Current to optical output behaviour ... 38
Figure 3.7 On-Off intensity modulation of a LED forward current ... 39
Figure 3.8 Block diagram of OOK system architecture ... 40
Figure 3.9 4-PPM waveform ... 41
Figure 3.10 DPIM waveform ... 43
Figure 4.1 Night sky irradiance scenario on I2C VLC link ... 50
Figure 4.2 SNR vs LED forward current ... 53
Figure 4.3 SNR with varying horizontal distance ... 53
Figure 4.4 bit-error-rate performance for different modulation schemes over horizontal distance ... 54
Figure 4.5 Street lamp distribution along the road ... 55
Figure 4.6 Measurement setup grid ... 55
Figure 4.7 Street lamps illuminance (lux) over the road ... 56
Figure 4.8 Signal-to-noise ratio along the road for different bandwidths, with artificial light interference ... 57
Figure 4.9 Slot error rate for DH-PIM1,2, DPIM, PPM and OOK-NRZ ... 58
Figure 4.10 Daytime high irradiance scenario on I2C VLC link ... 58
Figure 4.11 Signal to noise ratio vs LED forward current. ... 61
Figure 4.12 BER for OOK-NRZ modulation with different data rates ... 61
Figure 4.13 Optical Short-pass Filters Transmittance ... 62
Figure 4.14 bit error rate comparison with different optical filters, with PD exposed to high optical solar radiation ... 63
Figure 5.1 C2C scenario ... 65
Figure 5.2 Illustration showing the right-hand traffic, asymmetrical low beam pattern characterized by an extended visual range along the right side lane of the road [34] ... 66
Figure 5.3 Relative spectral power distribution of a White LED lamp (CMT1922, from Cree) ... 67
Figure 5.4 Radiant diagram for headlamp light distribution ... 68
Figure 5.5 Illuminance on a vertical plane, at 7 m distance ... 69
Figure 5.6 C2C communication link with direct path and diffuse reflected path ... 71
Figure 5.7 Diffuse Lambertian reflection on the road surface for an area element ... 71
Figure 5.8 Average received power over the roadway service area ... 72
Figure 5.9 Channel impulse response example ... 73
Figure 5.10 DRMS – of the service area; link connection between cars ... 75
Figure 5.11 Slot error rate for DH-PIM2 and DPIM, compared to bit error rate of PPM and OOK. Under moon optical background irradiance. ... 77
Figure 5.12 Slot error rate for DH-PIM2 and DPIM, compared to bit error rate of PPM and OOK. Under street lamp illuminance. ... 78
Figure 5.13 Slot error rate for DH-PIM2 and DPIM, compared to bit error rate of PPM and OOK. Under solar background irradiance. ... 79
List of Tables
Table 2-1 LED optical characteristics found on datasheet’s ... 8
Table 2-2 Luminous intensities (I) for traffic signal lights over the reference axis. ... 17
Table 2-3 Simulation Parameters ... 22
Table 3-1 Different LEDs modulation bandwidth ... 32
Table 4-1 Full Moon Earth’s surface illuminance ... 51
Table 4-2 Measured Solar illuminance at ground level... 59
Acronyms
AM Amplitude modulation
AWGN Additive white gaussian noise
APD Avalanche photodetector
BER Bit error rate
C2C Car to car
CIR Channel impulse response
DD Direct detection
DH-PIM Dual header pulse interval modulation
DPIM Differential pulse interval modulation
DRMS Delay root mean square
EM Electromagnetic
FM Frequency modulation
FOV Field of view
I2C Infrastructure to car
IM Impulse modulation
IR infrared
ISI Inter symbol interference
LAN Local area network
LD Laser diode
LED Light emitting diode
LOS Line of sight
ODAC Optical digital to analogue converter
OWC Optical wireless communications
PAPR Peak to average power ratio
PD Photodetector
PER Packet error rate
PM Phase modulation
PSD Power spectral distribution
PPM Pulse position modulation
RF Radio frequency
SER Symbol error rate
SNR Signal to noise ratio
Sr Steradian
1. Chapter 1
1.1 Introduction
In today’s world the voracity for information and entertainment is straining the available user bandwidth regarding the individual consume, where common people dry out the available RF spectrum [1] in telecommunications networks causing blockage and poor access to services, especially in crowded events. In VLC (visible light communication) the available spectrum is between 430 THz and 770 THz, as showed in fig. 1.1 [2] which comprises a very wide slice of the electromagnetic spectrum, offering freedom in terms of spectrum regulation and bandwidth availability.
The use of optical sources to convey information was always embedded in human nature. Ancient tribes used bonfires to signal an event that could be spotted from afar, it was especially useful when scouts wanted to inform faster their side of an oncoming enemy army or raiding party. In the XVIII century a French inventor and scientist Claude Chappe, invented the optical telegraph [3], a system of towers spread throughout France, these towers had two arms on each side and a cross-arm connecting each other, as seen in figure 1.1, enabling a large number of discrete and different positions of the structure, so encoded messages could be relayed from tower to tower. Napoleon used this system to transmit and receive information through his empire.There is even
modulating a beam of light, it was capable of reaching a distance up to 213 meters. These were the ancestors of VLC in which they represent a natural human pursuit in the development of such technologies.
VLC has been gaining interest of researchers and companies in the last two decades. It is only until recently that we have a disseminated resource that can modulate light to achieve meaningful data rates. This was due to continuous investment and development in the evolution of LED (light emitting diode) technology. Where it can be point out some factors such as the increase in the electrical-optical efficiency, a fairly linear relationship in terms of optical power vs current and also the high flickering capabilities in impulse modulation (IM) of the transmitted optical power in this kind of semiconductor enables the use of VLC systems. So it can be said that LEDs can be perceived as a high speed wireless transmitter, as so it is bandwidth can achieve values in the range of tenths MHz [5].
This work aims to use VLC in ITS (intelligent transportation systems) applications. ITS is a new trend trying to face arising problems in the growing road traffic. ITS is divided in three major architectures, finding different applications branches [6]:
• Infrastructure to car (I2C): intelligent traffic control, traffic information, traffic monitoring, emergency vehicle presence warning.
• Car to car (C2C): collision warning systems, cooperative traffic, situation awareness. • Car to infrastructure (C2I): traffic management for priority vehicles, electronic tolling. Currently the car and infrastructure lighting devices are being widely updated with LED technology, enabling VLC to play a major role in future ITS architectures.
1.2 Motivation
Since the beginning of the millennium, the percentage of fatalities related to car accidents that occur in city roads has increased in more than half of the IRTAD [7] (International Traffic Safety Data and Analysis Group) countries. In these group countries like Portugal, Greece and Korea present a significant increasing mortality associated to urban road traffic accidents. The Global status report [8] refers that the total number of road traffic deaths worldwide has reached the number of 1.25 million people in 2013. This statistic shows that there is a great necessity to decrease the human casualties related to this problem, ideally bringing them to zero. In order to be closer to this goal, many approaches have been considered, such as more restricted regulations on speed limits, seat belt use, policies to reduce drink-driving cases and better active restraint systems [8]. Intelligent transportation systems can be part of a solution designed to create a communication bridge between road entities in a traffic scenario, specifically cars and road associated infrastructure.
VLC can play a major role in ITS applications, providing several advantages over RF (radio frequency)[1]:
• Low complexity and cost: LEDs are already present in many possible application scenarios, such as car headlamps, signal traffic lights and street lights.
• High positioning: highly directional line of sight (LOS) propagation increases positioning accuracy.
• Improved link quality: high directional communication mitigates traffic congestion, particularly during rush hours.
• Security: offers higher security due to LOS propagation characteristics of VLC.
• Camera based communications: more complex receiver system, that allows spatially separation of different emitters.
• No spectrum regulation or licencing.
In VIDAS [9] one of the proposed scenarios for VLC, one of which is in scope in this work, was to develop a communication link between a traffic light signal and oncoming cars (I2C). One great remark of this project is to reutilize existing infrastructures such as the traffic light maintaining its original purpose and making it an emitter for visible light encoded data. The same can be said about the other scenario under scope C2C (car to car), the emitter would be implemented in the car’s
transmit traffic congestions in the nearby area, weather implications and accidents notices, and so on… This would be possible by integrating the visible light wireless information system with pre-existing traffic management systems like GERTRUDE[10]system in Lisbon. The possibilities for a C2C communication link can aid in the relay of information between cars of their relative positions, even automatically activate emergency brakes to avoid collision if the car ahead makes an unexpected stop. Even in a near future with the development of self-driving cars, C2C communication with OWC can provide an important tool in autonomous vehicles. So in a near future a LED lit car may not just be communicating with other nearby cars, but integrated in a complex mesh of devices in a city street grid [11].
1.2.1 VLC Applications
Everyday new ideas for applications in VLC are in development, where these are inherent to some specific positive proprieties in VLC which include decent bandwidth, light does not present any danger to the human body, low power consumption being a technology to be implemented in lightning devices, access security and no interference with electromagnetic sensitive equipment. Different applications for VLC which can include [12][13]:
• Hospitals – near electronically sensitive equipment like MRI scanner and operating theatres is undesirable to have Wi-Fi or mobile phones so it would be desirable to use VLC networks in these areas.
• Information displaying billboards – publicity sign and billboards often made from LED arrays where information can be modulated and received by a camera or other sensor in a user’s smartphone. This type of marketing information augmentation could be used in various locations such as airports, museums, hospitals or any other kind of business.
• Wireless local area networks – a unique trait for VLC in data transmission, is that data cannot be accessed by any device not in the same room as the systems transceiver. Unlike radio-based systems, VLC systems are limited by physical barriers like walls, which visible light cannot pass through. This makes necessary for the user to be present and in the same room as the system, providing an additional security feature.
different lights as an emitter group it would be possible to locate receivers within a room using trilateration techniques. Applying receivers’ tags to people and equipment would permit to discover their location.
1.3 Objectives
The main objectives in this work is to study and simulate the scenarios under scope (I2C, C2C), originally based on the VIDAS project [9], to better understand the necessities, impairments in this kind of wireless communication and trying to explore alternatives to the already developed VLC system in VIDAS. The constraints can be channel bandwidth, VLC’s systems optical propagation, sources of optical noise interference and data encoding. In order to try to overcome or minimize the effects of limitations in VLC scenarios.
1.4 Dissertation structure
This dissertation includes the current chapter with the motivation, context and some VLC applications. In the next chapters there will be present the following topics by order:
• Study the emitter and receiver characteristics.
• Development of an I2C MATLAB scenario, for LOS optical propagation. • Characterization of optical noise sources.
• System bandwidth limitations on the LED emitter.
• Proposed VLC modulation schemes analysis and performance comparison. • Performance comparison of different modulation schemes with different
optical noise ambient scenarios.
• Development of a C2C scenario with LOS and non-LOS diffuse link, with characterization of the channel.
2. Chapter 2
2.1 System Model Description
The purpose of designing a communication system is to deliver information from the emitter to the receiver without any difference from the source. There are many factors at play such as imperfections, receiver and emitter layout, channel noise and interference, signal attenuation or bad signal coverage, the received data is susceptible to error or even non-existent due to link loss. To be able to design a VLC system is necessary to have in mind the full limitations and different elements in an optical wireless channel, to do this is required a mathematical model capable of describing the optical propagation as well as the emitter and receiver behaviour. In this chapter it is addressed channel modelling for VLC systems focused on the LOS link, which in an outdoor application severely restricts the data communication range. A typical I2C scenario is represented in fig. 2.1, where the signal traffic light (Tx) transmits an optical signal to the red car (Rx).
To begin modelling the I2C scenario is necessary to analyse two basic characteristics of the LED emitter, the luminous intensity and the radiated optical power, these two metrics are directly correlated – being the first a measure of how a human being perceives the intensity of light and the second the energetic measure of the power radiated (photometric and radiometric relation
by proprieties of the LED itself and user safety regulations, therefore it also has a direct impact in the range of the data communication link in free space. Other major impairments arise with the presence of optical interference sources, some from natural origin (sunlight, moonlight), while others are artificial light (street lightning, other vehicles, etc.). The effect of this optical noise will have to be introduced in the receiver model, which normally is a photodetector diode [14][15][1].
2.2 Channel Modelling
The communication channel can be broken down into 3 distinct elements, the transmitter, the receiver and the physical space between them. The channel model revolves around the characterization of these three elements and their properties. Due to similarities of the communication medium between VLC and IR (infrared) systems, the knowledge gathered from extensive research in this predecessor’s systems (IR) can be adapted to produce mathematical models for VLC [16].
2.2.1 Emitter technical characteristics
As discussed before the usage and development of LEDs has brought the necessary viability for high bandwidth VLC systems, their fast switching capabilities and electro-optical efficiency factors in this account. Upon researching several manufacturer’s datasheet’s for different type LEDs, it was identified the necessary type of technical parameters to define the emitter in the simulation scenarios.
Table 2-1 LED optical characteristics found on datasheet’s
Manufacturer Model Colour Peak Wavelength Optical Spectral Width Luminous Intensity IV[cd] @ IF [mA] Half power angle Vishay TLCR5200 Red 616 nm 11 nm 4 cd @ 50mA 15°
TLCTG5200 True Green
520 nm 37 nm 2 cd @ 30mA 15°
OSRAM LR W5SM Red 628 nm 18 nm 22 @ 400mA 60°
LER Q8WP Red 632 nm 18 nm 50 @ 1400mA 60°
Avago HLMP- EG1A-Z10DD Red 626 nm 28 nm 12-21 cd @ 20mA 15° HLMP- CM1G-350DD Green 525 nm 20 nm 27-59 cd @ 20mA 15° HLMP-EL1A-Z1LDD Amber 590 nm 13 nm 12-21 cd @20mA 15°
The light emitting diode is a semiconductor P-N junction, where under enough electrical potential applied to its terminals a current flows and electrical carriers (electrons) recombine with holes. In this recombination process is usually generated a photon, but sometimes non-radiative recombination can occur, and the energy is converted to vibrational energy (phonons) and thus the electric energy is converted to heat. The electro-optical efficiency (η) tends to decrease with increasing driving current, when high carrier injection occurs non-radiative recombination increases and vibrational heat is generated [17]. This deficiency in the LED electro-optical performance causes non-linear effects in the communication channel distorting the information sent, but for low injection current the behaviour is approximately linear. When evaluating the electro optical efficiency of a red LED [18], a linear approximation for the efficiency can be used as displayed in figure 2.2.
As it is displayed the luminous intensity (Iv) efficiency drops with the increase current through
the LED. In the proposed interval it can be expect a relation between the optical intensity and forward current described by the electro-optical efficiency (ƞv), relative to a single LED:
𝐼𝑣
𝐼𝐿𝐸𝐷 = 𝜂𝑣[ cd
A] ( 2.1 )
The ratio between light output intensity from a single LED and the injection current, can also be modelled as [1]:
𝜂𝑣 = ℎ𝑣
𝑞 𝑛𝑖𝑛𝑡𝑛𝑒𝑥 ( 2.2 )
Where the electro optical efficiency depends on factors such as the internal quantum efficiency (nint), which measures the effectiveness ratio of carrier photon generator to total number
of carriers passing through the junction, the external efficiency (next) represents the ratio of photons
externally emitted by the device and not trapped or absorbed by imperfections in the LED casing, h is Planck constant, v the frequency of the generated photons and q the electron charge constant.
As it can be observed in fig. 2.3, LED as a light source is not monochromatic, the red LED spectral power distribution has its peak around 626 nm with optical spectral width around 28 nm. The relations established above in this subtopic are not wavelength dependent but based on photometric measures, like luminous intensity.
Photometry measures are analogous to radiometric units, as photometry is a branch of the wider field of radiometry. Radiometry can be considered as an independent entity capable of measuring the energy of electromagnetic radiation, while photometry considers the detector entity sensitivity to visible radiation, more specifically the sensitivity of the human visual system. Radiometry and photometry are also concerned with the optical power radiation for a given geometry of propagation. The photometric units analogous to the radiant flux ɸe, radiant intensity
Ie, irradiance Ee and radiance Le are the luminous flux ɸv, the luminous intensity Iv, the illuminance
Ev and the luminance Lv [1].
Normally LED manufactures express their optical characteristics in photometric units, but the relation between the radiometry and photometry can be given by:
ϕ𝑣= 𝐾𝑚∫ 𝜙𝑒(𝜆). 𝑉(𝜆)𝑑𝜆 780𝑛𝑚
380𝑛𝑚 ( 2.3 )
Where, V(λ) represents the normalized spectral sensitivity curve of the human visual system under daytime conditions, as seen in fig 2.3, the Km = 683 lm/W, is a constant establishing the radiometric to photometric sensitivity peak of the human physiological measure system. All the other photometric quantities are related to the weighted integral relationship of their corresponding radiometric units. To obtain the wavelength dependant optical power ɸe(λ),
normally is used a powerful computing tool like MATLAB to invert the weighted sum interval on the visible slice of the spectrum.
2.2.2 Emitter Model
Normally LEDs optical power radiation is approximated by modelling the emitter as a single point source with a Lambertian radiation pattern, given by:
𝐼(ɸ) = 𝐼0. 𝑐𝑜𝑠𝑚(ɸ) ( 2.4 )
The radiant intensity (W/sr) has its maximum when the viewing angle (φ) between the propagated light intensity and the emitter normal axis is 0 degrees and m is the mode number of the Lambertian emission that defines the directivity of the source. In figure 2.4, it shows a radiant intensity pattern for an ideal Lambertian source with m=1, equivalent to an LED with half power angle of 60° degrees.
𝑚 = − 𝑙𝑛 2
𝑙𝑛(𝑐𝑜𝑠 ɸ1/2) ( 2.5 )
Where ɸ1/2 is the half power angle of the LED dependent on geometric and encapsulation
conditions of the semiconductor. The LED emitter is modelled using a generalized Lambertian radiation pattern [1], where the radiant intensity RE(ɸ,m) is dependent on the viewing angle and
the directivity of the source and assuming Pt as the total power emitted by it.
𝑅𝐸(𝜙, 𝑚) = 𝑚+1 2𝜋 𝑃𝑡𝑐𝑜𝑠 𝑚(𝜙); ɸ ∈ [−𝜋 2; 𝜋 2] ( 2.6 )
The 𝑚+12𝜋 coefficient ensures the integration of the radiant source power Pt over the directivity
lobe of the emitter, to result in the radiant intensity emitted dependent on the emission angle as showed in equation 2.4. Which in this case of a Lambertian source of visible radiation establishes the relation between the radiant power and the radiant intensity of the LED.
From equation2.5, the directivity of the emitted power is dependant to the half power angle (hpa), the viewing angle at which 50% of the main emitted radiant intensity falls to its half value. Narrow hpa causes LEDs to have more axial directivity of the propagated power, but the radial intensity falls off more quickly. Figure 2.5, shows different radiant intensity patterns for different hpa values, corresponding to 20°, 30°, 45° and 60° degrees.
2.2.3 Traffic Light Emitter Matrix
LED signal traffic lights use high brightness LEDs connected in a combination of series and parallel to build a LED matrix, adding an optical lens is also necessary to have a uniform light distribution regarding the normal reference axis of the light emitter. In this subtopic is presented a circular LED arrangement and the resulting luminous intensity generated by the LED traffic light, considering the requirements in regulations for signal traffic lights.
The simulated LED matrix display (figure 2.6), has the following characteristics: • 200 LEDs over 11 co-centric rings;
• Equal inner-ring LED spacing and equal inter-ring spacing;
The ring pattern of LEDs placed in co-centric circumferences, as shown in fig. 2.6, form a source of multiple LED Lambertian emitters. The models in this subchapter consider the emitting matrix as the centre of origin, in the Cartesian coordinate system (x,y,z), for simplification purposes the matrix emitting plane is parallel to the XoY plane.
Taking into consideration only one LED, for example placed in the centre of the emitting plane with coordinates (x0, y0, 0), and a heading of the emitting source, 𝑛⃗⃗⃗⃗ . It is possible to calculate the 𝑠
emitting angle, for a determined target at any point (x, y, z), given by: cos(𝜑) = 𝑛⃗⃗⃗⃗ .𝑑 𝑠
‖𝑛⃗⃗⃗⃗ ‖‖𝑑 ‖𝑠 ( 2.7 )
Where the angle is obtained from the dot product between the distance vector, 𝑑 , and the heading of the source. With the norm of the heading ‖𝑛⃗⃗⃗⃗ ‖ = 1 (𝑛𝑠 ⃗⃗⃗⃗ = (0, 0 ,1)), the norm of the 𝑠 distance is:
‖𝑑 ‖ = √(𝑥 − 𝑥0)2+ (𝑦 − 𝑦0)2+ (𝑧)2 ( 2.8 ) Given this geometrical setup, in fig. 2.7, the irradiance reaching on the target point (x, y, z) is given as [14]: 𝐸𝑒(𝑥, 𝑦, 𝑧) = 𝑚+1 2𝜋 ɸ𝑒 𝑐𝑜𝑠𝑚(𝜑) ‖𝑑 ‖2 ( 2.9 )
That can become,
𝐸𝑒(𝑥, 𝑦, 𝑧) = 𝑚+1
2𝜋 ɸ𝑒 𝑧𝑚
‖𝑑 ‖2‖𝑑 ‖𝑚 ( 2.10 )
Where, ɸ𝑒 is the radiant flux of the emitter. Finally, the irradiance can be rewritten as: 𝐸𝑒(𝑥, 𝑦, 𝑧) = 𝑚+1 2𝜋 ɸ𝑒 𝑧𝑚 [(𝑥−𝑥0)2+(𝑦−𝑦0)2+𝑧2]( 𝑚+2 2 ) ( 2.11 )
An example of an inner LED co-centric ring is showed in fig. 2.7, placed in the XoY plane with each LED having a Cartesian position of (xi, yi, 0). The distribution of multiple LEDs over the
co-centric rings, can be given by:
{ 𝑥𝑖 = 𝑟𝑘. cos ( 2𝜋 𝑁𝑟(𝑘)𝑖 + 𝜃) 𝑦𝑖 = 𝑟𝑘. sin ( 2𝜋 𝑁𝑟(𝑘)𝑖 + 𝜃) ( 2.12 )
Where i is the index of a LED on a ring, since each ring has a certain number of LEDs, k is the ring index as there will be multiple co-centric rings in the matrix.
Considering now multiple points of radiant optical power, multiple LEDs in a single ring, with Nk being the number of LEDs on a ring and M the number of multiple co-centric rings. The total
irradiance can be given by: 𝐸𝑒(𝑥, 𝑦, 𝑧) = 𝑚+1 2𝜋 ɸ𝑒∑ ∑ 𝑧𝑚 [(𝑥−𝑥𝑖)2+(𝑦−𝑦𝑖)2+𝑧2]( 𝑚+2 2 ) 𝑁𝑘 𝑖=1 𝑀 𝑘=0 ( 2.13 )
Index k is a part of the index i, for the representation of the discrete LED coordinates (xi, yi,
0) on the XoY plane.
When the target point goes further and further away from the emitting matrix, the z coordinate increases more than the others as the LOS analysis is made mostly on the longitudinal view from the signal traffic light. What happens is that the emitting matrix of discrete points can be approximated by a single point of emission based on the central point of the circular array, due to the xi, yi LED discrete coordinates magnitude, is low in comparison with the target point coordinates
magnitude. As the irradiance can be approximated by:
𝐸𝑒(𝑥, 𝑦, 𝑧) = 𝑚+1 2𝜋 ɸ𝑒𝑁𝐿𝐸𝐷𝑆 𝑧𝑚 [(𝑥−𝑥0)2+(𝑦−𝑦0)2+𝑧2]( 𝑚+2 2 ) ( 2.14 )
Following the equations of a Lambertian radiance power source model, as described before, and using the linear approximation (obtained from the LED HLMP-EG1A-Z10DD datasheet from Avago) from the relationship between a single LED luminous intensity vs forward current, it is possible to obtain the propagated radiant intensity from the LED matrix on the normal axis. The irradiance obtained at 1m from the normal axis of the matrix is equivalent to the radiant intensity.
In fig. 2.6 and 2.8, it is showed the proposed LEDs distribution and the resulting luminous intensity vs driving current.
The simulations provide a LED distribution matrix for computed standard luminous intensity value dependent on the LED driving current, in order to evaluate the compliance with the European regulatory norms and requirements for traffic control equipment.
Table 2-2 Luminous intensities (I) for traffic signal lights over the reference axis.
Performance level 1 2 3
Imin 100 cd 200 cd 400 cd
Imax class1 400 cd 800 cd 1000 cd
Imax class 2 1100 cd 2000 cd 2500 cd
Therefore, the emitter physical and optical characteristics comply with:
• Diameter for signal traffic lights: round signal light with a nominal diameter of 200mm. • Performance level class: 2/2
2.2.4 Receiver Model
In VLC systems there are different types of devices capable of detecting optical signals and represent them in the electronic domain. These range from digital cameras to photodiodes; digital cameras have usually a low acquiring frame rate and are used for multiple access systems or for communications systems that employ multilevel ODAC codification; in case of photodiodes there are two options avalanche photodetector (APD) and PIN photodiodes, where the latter is preferable for outdoor communications link due to the presence of high optical noise sources as APD can greatly increase the noise component with their intrinsically gain effect.
Signal reception model is of paramount importance in the description of the received signal by the optical-electronical front end. Considering the use of a PIN photodiode, this device can sense the optical power reaching its photo detection active area over a wide spectral range, usually where the visible light is included. This process is described the PD spectral responsivity, where the incident spectral power is integrated and produces a photo generated electrical current, this is usually a good measure of the device performance [1].
Photodiodes based on SI junctions usually have a similar response to that of the example on fig. 2.9, based on the FDS100 from Thorlabs. Despite having a full coverage on the visible spectrum, the responsivity maximum occurs in the near IF part, where only optical noise sources will contribute with high spectral power density in this region. Due to a lack of a better choice in the photosensitive diodes industry with a better spectral response in the exclusive visible range. To solve this problem, the system could include an optical short-pass filter, like the 700SP which cuts off some of the IR interference component on the receiver, however the filter alone costs around 30 €.
Responsivity is defined as the ratio between the incident optical power and the generated photocurrent, is dependent on the quantum efficiency (η) of the sensitive material as represented in equation 2.15 [1], where q is the electron charge, λ is the wavelength, h is the Planck constant and c the speed of light.
Ɍ(𝜆) =𝐼𝑃𝑝ℎ 𝑜 = 𝜂
𝑞𝜆
ℎ𝑐 ( 2.15 )
𝐴𝑒𝑓𝑓(𝜎) = {𝐴0 ; |𝜎| ≥ 𝐹𝑂𝑉𝑑. 𝑐𝑜𝑠(𝜎); |𝜎| < 𝐹𝑂𝑉 ( 2.16 )
The model results in eq. 2.16, which provides the effective signal collecting area of the photo receiver, which depends on physical/geometric parameters such as the PIN active photosensitive area (Ad) and the angle of incidence (σ) of the incoming optical radiation which depends on the
relative position of the receiver to the emitter. Photodetectors have a limited FOV, the incident angle of radiation must be within the limits of the detector specified by the precited parameter.
Narrowing the receiver FOV and consequently increasing its directivity towards the emitter can improve signal reception and cutting off some undesired optical background interference noise, but that would require to the emitter and receiver to be aligned, which can happen in this I2C scenario but will not be normal occurrence. To complete this model is further necessary to provide some additional information of the receiver, its position and vertical inclination respect to the normal of the road and as shown in figure 2.11.
2.2.5 I2C LOS Model Description
A simplified model for I2C communication is illustrated in fig. 2.1. After pre-processing valuable traffic data will be transmitted to oncoming cars from the signal traffic light, modulating the visible light intensity of the LED emitter. The LOS model approximates the power that reaches the receiver based on a photodetector and an amplifying stage capable of processing optical information and amplifying the signal. The parameters and the model specifications for the emitter and the receiver can be found in table 2-3, where the red traffic light must be housed on the top of the signal head, whilst standing poles of the traffic signal must be placed near the roadway, with a height comprehended between 2m to 3.5m, in accordance with the regulations decree [19].
In this model, it is assumed a traffic light position in a 3D scenario with Cartesian coordinates that define its location on the origin (0,0,hE) and a variable position of the receiver (xr,yr,hR) but with
a fixed height, where X is the distance along the roadway width (in accordance with urban main roadways municipality regulations that specify a minimum roadway width of 7 m) and Y the distance along the road length (the considered or under evaluation service range). One of the major loss factors, which will be included in the LOS path gain model, is the distance between the emitter and receiver represented by the term d and the produced gain has an inverse square relation with the distance. This term is obtained by:
𝑑 = (𝑥𝑟, 𝑦𝑟, ℎ𝑅− ℎ𝐸) ( 2.17 )
and,
‖𝑑 ‖ = √𝑥𝑟2+ 𝑦𝑟2+ (ℎ𝐸− ℎ𝑅)2 ( 2.18 )
To obtain the emitting angle from the source and the incident angle on the receiver is necessary to define the heading of the source (EN) and the heading of the receiver (RN), these values
are set with the horizontal inclination and vertical inclination respectively, on table 2-3. Recurring to the internal product definition between vectors to obtain the transmission angle (ɸ) and the incident angle (σ) can be respectively defined by:
𝜙 = cos−1 (𝐸⃗⃗⃗⃗⃗⃗ .𝑑 )𝑁
𝜎 = cos−1 −(𝑅⃗⃗⃗⃗⃗⃗ .𝑑 )𝑁
‖𝑑 ‖‖𝑅⃗⃗⃗⃗⃗⃗ ‖𝑁 ( 2.20 )
Where the radiant intensity is given by [15]: 𝑅𝑖(𝜙) = 𝑃𝑡.
(𝑚+1) 2𝜋 . 𝑐𝑜𝑠
𝑚(𝜙) ( 2.21 )
These values can change when considering a moving receiver (xr, yr). Whereas EN and RN are
the emitter and receiver normal heading with unitary norm. The emitter receiver distance, or signal traffic light to car distance is a big factor when calculating the channel loss, and consequently the signal decay increases further is the distance of the receiver in the evaluation simulation range.
Table 2-3 Simulation Parameters
Emitter simulation parameters Receiver simulation parameters Description Notation Value Description Notation Value
Emitter half angle hpa 15° FOV of the PD FOV 65°
Luminous intensity Iv 200-2000 cd Physical area of PD Ad 13 mm2
Wavelength peak λpeak 626 nm Height hR 1 m
Spectral Bandwidth Δλ 28 nm Vertical inclination θ 80°
Horizontal inclination φ 5°
Responsivity Ɍ Fig 2.8
2.2.6 Optical Power Path- I2C
Optical signals propagate to the receiver along two distinct conditions, one is due to the direct link between the emitter and receiver, and the other due to reflections on nearby surfaces, but as the distances involved are at least one order higher than of those indoor VLC systems, where reflections are more abundant and considerable, (the only guaranteed reflecting surface on I2C communications being the road surface) these contributions are considered insignificant face to the direct LOS optical propagation path. So this model approximates the path loss with the LOS propagation model based on the work of Ghassemlooy [15] and the signal LOS path component only suffers a linear delay between transceivers.
In the configuration represented in fig. 2.1, the radiant optical power (Watts) collected by the photodiode depends on the path attenuation factor between the traffic light and the receiver on the vehicle, the receiver model, the transmitted optical power and the relative emitter/receiver orientations. Hence the incident transmitted power is given by:
𝑃𝑟 = 𝐻𝐿𝑂𝑆. 𝑃𝑡 ( 2.22 )
And the complete LOS channel DC gain, depends on the transmission angle, the incident angle and the emitter receiver distance:
𝑯𝑳𝑶𝑺(𝑑, 𝜙, 𝜎) = 𝑚+1 2𝜋𝑑2. 𝑐𝑜𝑠 𝑚(𝜙). 𝑐𝑜𝑠(𝜎) . 𝐴 𝐷. ℛ. 𝑟𝑒𝑐𝑡 ( 𝜎 𝐹𝑂𝑉) ( 2.23 )
In fig. 2.12, there is represented the overall power collected by photodetector over the road surface with a fixed height (hR). This can be extrapolated from a function of the emitter spectral
power of a red LED integrated over the spectral responsivity range of the receiver. 𝑃𝑟𝑥(𝑥𝑟, 𝑦𝑟, ℎ𝑅) = 𝐻𝐿𝑂𝑆. ∫ Ɍ(𝜆). 𝑃𝑡(𝜆)𝑑𝜆
𝜆𝑓
𝜆𝑖 ( 2.24 )
The dimensional results present over the channel show a longitudinal path loss due to the distance increase and the lateral loss is due to the high directivity of the source beam but this characteristic of the emitter helps compansate the longitudinal loss. The high directivity evidenced in this case scenario is due to the low hpa of the emitter (15°), compensatting the link performance
In figure 2.12, it can be seen a high received power loss near the beginning of the road length and even more when reaching the edge of the roadway, this is caused by the receiver being outside the LED’s irradiance angle and/or the incident optical light being outside the receiver FOV. The magnitude of the received signal implies a necessary gain on the receiver side in order to recover the transmitted information, but this is not the only problem because there is also the necessity to consider and simulate the induced noise introduced by external optical sources presents on the transmission channel.
2.3 Channel Noise Sources
Outdoor VLC systems receivers are subject to intense ambient light. Having different optical background sources, depending on many factors and from all kinds of different visible light sources. The optical light sources range from artificial to natural sources of light, such as the sun, night sky radiation, street illumination, other vehicles, advertising signs and so on… These sources will generate noise at the receiver, which in turn will add to the optical signal at the receiver input, degrading the signal-to-noise ratio and therefore will be referred as optical background noise sources.
Natural light sources produce a uniform constant interference over the VLC channel, these normally can be originated by the sun and moonlight. At night the natural light irradiance main source is due to the moon’s reflection of sunlight. During the day sunlight can present a constant irradiance level incident the earth’s surface or considering the channel evaluation area, if there are shadowing elements present (Ex.: clouds or tree’s) there will be slow variations of the irradiance intensity with time.
Artificial lights exhibit large intensity variations in time which creates an interfering signal at the receiver. This is not classified as noise because the interfering signal has a deterministic behaviour normally associated with lower band electric frequencies that can be filtered at the receiver stage. But nevertheless, artificial light also produces a DC component at the photodiode, which will generate noise.
2.3.1 Electronics Noise
Noise generated at the receiver consists of two primary components, thermal noise and shot-noise. Thermal noise is a product of thermally induce variations in resistive components. The noise voltage variance is given by:
𝑉𝑡ℎ2 = 4𝐾𝐵𝑇𝑅 ( 2.25 )
Where KB is the Boltzmann constant, Ͳ is the absolute temperature in Kelvin, B the observation
bandwidth and R the load resistance at the receiver.
Optical background irradiance sources cause shot-noise at the receiver input stage. The presence of ambient light sources, with different spectral distributions and an overall spread presence in the photodetector spectral sensitivity range, contribute to noise. The average incident optical power generates a continuous photocurrent at the receiver stage, which causes Shot-noise (n(t)) to rise. Shot- noise is associated with the passage of carriers through a potential barrier, this means that any photocurrent in the photodiode will have associated noise. Shot-noise is described by a Gaussian distribution with zero mean and a white power spectral distribution.
Where the spectral density of quantum shot-noise associated with the IDC current flowing
𝑁𝑜 = 2𝑞𝐼𝑏 ( 𝐴2
𝐻𝑧) ( 2.26 )
Where q is the charge of an electron and Ib is the total DC photocurrent in the receiver. The
total noise variance is then given by:
𝜎𝑛2= 2𝑞𝐼𝑏𝐵 ( 2.27 )
Where B is the baseband signal bandwidth.
In fig. 2.13, shows the spectral distribution for the Shot-noise being uniform throughout the spectrum, which is characteristic for white noise sources. The shot-noise presence can span up to the photodetector bandwidth, but a baseband signal spectral occupancy occupies a lower bandwidth, and this is the spectral occupancy range considered for the shot-noise calculations.
Note that when the optical signal power increases, with non-negative components the (IDCsig)
signal photocurrent generated increases, but the DC component contributing to shot-noise also rises:
𝐼𝑏= 𝐼𝐷𝐶𝑠𝑖𝑔+ 𝐼𝐵𝑔+ 𝐼𝑑𝑎𝑟𝑘 ( 2.28 )
Where IBg is the DC photocurrent generated by the background illuminance and Idark is a small
current that flows under no optical exposure. Shot-noise is the dominant noise source in OWC systems, the background current can be several times higher in magnitude when compared with the receiver signal, especially in the presence of direct sunlight [15].
Figure 2.13 Power spectrum of a constant photocurrent on a photodetector and the noise spectral distribution [20]
The noise present at the receiver stage can be modelled as AGWN (additive white gaussian noise), in information theory this type of noise is described by: i) Additive, the noise is randomly added to the signal; ii) White, has an even power distribution along the spectrum; iii) Gaussian, because it has a normal distribution in the time domain with zero mean value.
2.3.2 Signal and Noise
In the standards developed for communications systems there are important evaluation metrics between signal and noise (S/N or SNR). In digital communication this metric developed in the form of Eb/No, which is a normalized version of SNR. Where Eb is the energy per bit and No, as referred, is the noise power spectral density. Rewriting Eb and No, respectively as:
𝐸𝑏 = 𝑃𝑠𝑇𝑏 (𝑊. 𝑠) ( 2.29 )
𝑁𝑜 = 𝑁
𝐵 (𝑊. 𝑠) ( 2.30 )
Where Ps is the signal power, N is the noise power and B the bandwidth. Replacing the equivalent data rate in the bit duration and establishing the Eb/No relation.
𝐸𝑏 𝑁𝑜= 𝑃𝑠 𝑁( 𝐵 𝑅𝑏)⇔ 𝑆𝑁𝑅 = 𝐸𝑏 𝑁𝑜( 𝑅𝑏 𝐵) ( 2.31 )
The equation above shows Eb/No as a normalized version of the signal to noise ratio (SNR), one of the most important tools for evaluating a system performance in digital communications and also enables an evaluation for the bit error probability [21].
In optical wireless communication systems, considering only a LOS propagation of the optical signal, the Eb/No can be expanded into:
𝐸𝑏 𝑁𝑜=
𝑃𝑡𝐻𝐿𝑂𝑆(0)𝑇𝑏
𝑁𝑜 ( 2.32 )
Pt – transmitted optical power
HLOS(0) – channel DC response loss
The electrical SNR version, considering a majoring source of noise as Shot-noise and taking into consideration the electrical signal power generated at the optical receiver.
𝐸𝑏 𝑁𝑜=
(𝑃𝑡ℛ𝐻𝐿𝑂𝑆(0))2
2𝑞𝐼𝑏𝑅𝑏 ( 2.33 )
2.3.3 Black Body Radiation
The most common light source comes from the radiation emitted by thermally excited atoms, in the form of solid or gas bodies, as in the case of the Sun or incandescent lightbulbs. When considering a perfect emitter (black body) heated to an absolute temperature T, its spectral radiance can be empirically described by Planck’s law [22]:
𝑃(𝜆) = 2𝜋𝑐2ℎ
𝜆5(𝑒ℎ𝑐/𝜆𝑘𝑇−1) ( 2.34 )
According to this expression, the emitting body radiation spectrum presents a continuous behaviour (fig. 2.14) with an emittance peak at a certain wavelength (λm), which is inversely
proportional to the body’s temperature. This relation describes the radiance from infrared to ultraviolet spectrum.
• Sunlight: High irradiance from the sun at earth surface level, producing high levels of photocurrent at the receiver. The level of irradiance is not constant and highly dependent on external factors of a VLC channel. In order to produce a complete model we need to contemplate several factors; such as time of the year (which changes with the earth sun mean distance), geographical location (latitude) and the vertical altitude of the sun (which
also changes along the day)[22]. The sun radiates isotropic radiation with a spectrum approximated by its photosphere temperature 5780 Kelvin, the region generating light [23], on fig. 2.14 it can be observed the sunlight’s PSD upon the photodiode sensitivity range.
• Moonlight: Moon’s radiance on the earth’s surface caused by the reflection of the sun on this natural satellite is affected by a number of elements and variables [22] - the phase of the moon, the earth-moon distance variation, the differences in reflectance (albedo) surfaces in the moon, the relative altitude of the moon in relation to the horizon line and the weather conditions.
The moon as spectral radiance can be approximated by a blackbody radiation at 4100 Kelvin [24], in fig. 2.15 there is an example for the moonlight’s approximated spectrum from 0.38µm to 1.1 µm.
• Artificial lights:
o the incandescent light source produces light, in which this process involves resistively heating a tungsten filament inside the bulb with a current conducted through it. The tungsten filament heats up to around 2900 Kelvin when a steady state of operation occurs, and the modelled spectrum of radiation is a black body around that temperature. A lot of the radiated energy from these lightning sources also have a very strong component in the near IR of the spectrum in which visible light photodetectors have a strong responsivity.
Fig. 2.15, displays the normalized power spectral distributions of different light sources, with the spectral window scaled to the receiver sensitivity range. As it can be seen the Incandescent lamp as a stronger component in the infrared region of the spectrum due to their low efficiency and high radiation heat generation, whereas the sunlight radiation has a stronger component in the ultra-violet region.
3. Chapter 3
When developing an OWC system one of the most important factors is choosing an appropriate type of modulation technique to enhance the system’s performance, depending on the requirements for the link setup. Many of the existing VLC modulation techniques are based in radio and infrared technical backgrounds. The optical characteristics in a visible light emitter differ significantly from RF systems. In radio, proprieties of the carrier wave like amplitude, frequency and phase can be modulated by the transmitting data, whereas in VLC is only the intensity of the optical carrier, usually LED radiant power propagating in multiple directions which is a real nonnegative value, that can be modulated. LEDs are an incoherent light source, they emit light with multiple changes in phase and frequency [17]. The generation of light comes from the transition of an electron in high energy state to a lower state, the energy bandgap between these two levels is directly correlated to the frequency/wavelength of the emitted photon.
This chapter presents some basics around different types of visible light modulations and circuit approaches. Limitations and requirements that influence various aspects, such as bit rate vs bandwidth limitations, modelling error performance and power efficiency.
3.1 VLC Impairments
When designing the VLC system, there are some requirements over practical impairments to consider for the optical signal propagation. There are eye safety regulations [25] to comply with in terms of optical power. VLC emitters do not emit focused beams, such as LASER’s, which can cause possible damage to the eye. However, due to our “high sensitivity” to visible light there is an adequate limit on the mean optical power propagated by the emitter, but not as much as in IR communications.
Another concern in VLC systems is maintaining a “constant” light intensity, whistle varying the emitter intensity. Fast switching features, necessary for data communication are not always suitable for that purpose, since mean light intensity could change with the bit probability. To avoid this, there are some line codes such as Manchester, which maintain a mean DC value independently of the symbol to convey, thus levelling a constant light output at the emitter.
3.1.1 Channel Bandwidth Limitations
As referred before multipath dispersion will not be included for the I2C simulations, it is only considered the LOS propagation. Multipath dispersion will be a subject addressed in chapter 5 with the C2C scenario.
Signal bandwidth limitations can be from different elements in the communication system, like the receiver configuration or the emitting LEDs.
The emitting LEDs do not have an infinite bandwidth, this is one of the major limitations for high data rates in VLC systems. The recombination dynamic processes inside the semiconductor junction are limited by the minority carriers’ lifetime, which limits the time between the transitions of the LED on and off state. This process defines the optical bandwidth of the emitting device, which occurs when the parasitic junction capacitance starts pulling the driving current down, [17]. In table 3-1, there is a list of tested LEDs for their optical bandwidth, with the respective maker, model, colour and bandwidth in MHz.
Bandwidth measured with OOK modulation and LED response times [5][26]: Table 3-1 Different LEDs modulation bandwidth
Manufacturer Model Colour IF (max)
[mA] Optical Bandwidth [Mhz] Switching Time [ns] OSRAM LA W57B Amber 400 3.10 LY W57B Yellow 400 4.10 LV W5SG Green 500 8.70 OSLON SSL 80 White 800 3.0 VISHAY TLMK3100 Red 30 8.30 42.11 TLCR5200 50 7.50 46.63 TLCR5100 50 7.23 48.35 TLCR5800 50 7.50 46.63 TLCY5100 Yellow 50 5.89 59.38 TLCY5200 50 8.04 43.52
TLCY5800 50 5.89 59.38 TLCTG5100 Green 30 6.53 53.54 TLCTG5800 30 8.04 43.52 TLCTG5200 30 7.5 46.63 TLCB5100 Blue 30 11.75 29.76 TLCB5200 30 12.16 28.75 TLCW5100 White 30 2.40 TLMW310 20 2.40 Nitchia TLMW3100 20 2.40 145.94 NSPW500BS 30 2.57 136.20
Multicomp OVA-1033 Blue 30 9.9
OVA-1031 White 30 2.8
OVA-1031 White w/ blue filter 30 3.45
TT Electronics OVLFR3C7 Red 50 5-6
The results present in table 3-1, are a collection of different sources and work developed in Integrated Systems and Circuits group from IT Aveiro, where the last entry on the table was made from an experimental setup and measurement develop in this work.
VS(VNA) Vbias RE ILED Rin Vcc Vcc R1 R2 Cin
The experimental setup consisted in a transmitting LED at a distance of 2 cm to the receiving photodiode, the system bandwidth was measured by using a VNA (vector network analyser) where port 1 was connected to the emitter and port 2 was connected to the receiver. The bandwidth was obtained by getting S21 parameter magnitude. The adaptation circuits required for driving the LED and photodiode polarization, are represented by fig. 3.1 and fig. 3.2 respectively. The LED driving is accomplished by using a bipolar transistor, in a common emitter configuration with the LED in series with the transistor collector. The resistor on the emitter, RE, sets the current on the LED, given by:
𝐼𝐿𝐸𝐷= 𝐼𝑏𝑖𝑎𝑠+ 𝐼𝑠 ( 3.1 )
Where Ibias is the DC bias current and Is the signal current driving the LED both established by:
𝐼𝑏𝑖𝑎𝑠= 𝑉𝑐𝑐 𝑅2 𝑅1+𝑅2−𝑉𝑏𝑒 𝑅𝐸 ( 3.2 ) 𝐼𝑠= 𝑉𝑠−𝑉𝑏𝑒 𝑅𝐸 ( 3.3 )
The optical receiver is a reversed biased photodiode, with an input noise filter, implemented by R1 and C1. The bias voltage was set to 20V, which according to the datasheet provides a junction capacitance, Cj, of 24pF. The receiver bandwidth is given by:
20V
Vout
RL
R1
C1
𝐵𝑊 = 1
2𝜋𝐶𝑗𝑅𝐿 ( 3.4 )
With a load resistance, RL, of 50 ohms the implemented receiver bandwidth is approximately
132.6MHz. Since the expected LED bandwidth is around few MHz, this limitation has a small impact on overall system bandwidth.
The measurement results in fig 3.3, were made on VNA. The cut off frequency relation can be given by: |𝐻(𝑓𝑐)| = 𝐻(0) √2 ( 3.5 ) Being, 𝐻(𝑓) =𝑉𝑜𝑢𝑡(𝑓) 𝑉𝑖(𝑓) ( 3.6 )
Analysing the data obtained from the VNA measurements, the expected LED optical bandwidth was according to other research measurements on LEDs. Optical receiver bandwidth limitations are not of high concern, due to the existence of other architecture configurations with lower load impedances [27], increasing even further the available receiver bandwidth.
3.2 Digital Baseband Modulations
The purposed modulation schemes are included in this class, of baseband modulation, where the data/symbol transmission is not shifted to a high carrier frequency before the intensity modulation of the optical source. After modulation being applied to the transmitted data a line code can constrain even more the transmitted signal shape. Line codes usually provide some advantages which can range from clock recovery to facilitate synchronization, to capability of error detection and correction. The generated and transmitted signals in visible light applications after modulation have a strong DC component in their power spectral distribution analysis, this is due to optical signal generated, can only be considered bipolar with a DC offset, being its amplitude ranging only in positive values. All optical visible light modulations optical signals can be considered unipolar. Free Space channel DC bias Pt Data IDC
Modulator isignal ILED
Pinc Shot Noise Amplification Signal Processing Demodulation Ipd Recovered Data
Fig 3.4 represents a typical VLC system setup with main blocks for the emitter and receiver stages, in the emitter it is displayed the DC bias point applied to the emitting LEDs at which is added the modulated signal. The transmitted optical signal has a DC component based on the DC current bias point of polarization.
In on-off keying modulation with non-return to zero (OOK-NRZ) of a light source, the optical signal is in the form of a rectangular pulsed shape. In the time domain the rectangular pulse can be given by:
𝑥(𝑡) = 𝑟𝑒𝑐𝑡𝑇(𝑡) ( 3.7 )
The power spectral distribution of the rectangular function with pulse width Ͳ is given by the Fourier transform of the time domain signal and is approximated by [28]:
𝑋(𝑓) = 𝐴𝑇sin (𝜋𝑇𝑓)
𝜋𝑇𝑓 ( 3.8 )
Which is the sync function represented in figure 3.5, also in this figure is represented the rectangular pulse shape typical of OOK-NRZ modulation scheme. The data rate Rb (bits/s) is inversely
proportional to the pulse duration Ͳ and as the bandwidth occupancy is between the DC point up to the first null in the PSD of the transmitted signal, the bandwidth B is also inversely proportional to the pulse duration. Which gives the theoretical spectral power efficiency for OOK-NRZ to be limited to 1 (bit/s(Hz) and considering the LED pole, this can turn into a major limitation to data rate transmission in visible light systems.
